Generell discussion about the cube scripting language

mapc | 2011-12-16 12:57

RaZgRiZ | 2011-12-12 13:00

Cubescript is hard to read to those who are either too accustomed to reading everything with syntax highlighting or don't like cubescript in any way. Or when someone is truly horrible at doing things right. Apart from that however..

He may not be the first, or the second, or the third, what does it matter. For cubescript, it's practically the first command-line quick-connect script, with mine coming second but being SVN only. I do wonder however how you're doing to implement such a thing into the framework, is it based on cubescript at all? Or are you hoping to convert it to another language?

And i'm confident he will add the comments necessary. Would would go past thing chance for fame? ^_^
reply to 3starRaZgRiZ

mapc | 2011-12-12 17:46

(I switched from the gandalf acc to this one, sorry for the confusion).

Yes, apply 1 and 2 for me :D (but it is not that bad).
Sadly cbs is a very bad designed language: There is for instance only a global namespace and there is only strings. Even the lists are space separated, not escaped strings. (I'd like to give you a complete lists, but this would be one or two paragraphs of text, so not in this comment ;) ).

Yes, thats why I am thinking about integrating it :). However, command line again means a problem with the global namespace :(
The framework is entierely cubescript and aims at resolving some of the issues, nicer installation, plugin-based (no more hazzle with adding stuff to autoexec.cfg) and a little bit of namespacing:

(concatword $namespace $varname) = $value

Yup, hope so :)

Cheers,
mapc

srbs | 2011-12-14 04:46

@mapc
"I'd like to give you a complete lists, but this would be one or two paragraphs of text, so not in this comment ;)"
I'd love to see this list.

"Sadly cbs is a very bad designed language:"
Not every language NEEDS every feature known to the programming world. It is perfectly adequate for what it is designed for.

What isn't designed very well is, in fact, your "framework." For large projects like that, one should (as with all languages) first learn the features, quirks, and idioms of said language and design accordingly. And if my config.cfg file is some 500KB after using 0.9.4, I won't ever use it again, especially since every function of every script has variable names relating to the current plug-in

@Razgriz
"Would would go past thing chance for fame?"
wut?

srbs

mapc | 2011-12-15 12:48

@srbs

I do not criticize the lack of features in the language, but the way the language is designed (in other words: how you do, what you do with cbs).

Your last argument was an ad hominem argument (http://en.wikipedia.org/wiki/Ad_hominem) and therefore a rhetorical trick without being constructive at all. Please stop this, I am annoyed by this.
Your config file is that big, because the way configuration works (described below) and because of the lack of memory management (described also below). Big codebase => Big conf. This issue can not be resolved without having control of the configuration process. Sorry.
The framework is btw written with some of the issues below in mind. The usage of vars for every plugin for instance makes the whole framework much more secure.
The Points of criticism:

Everything is a string (partially lack of types)
Lists are space-seperated not-escaped lists
There are too many ways to generate and modify function-strings (e.g. macros, the three braces)
There are too many ways to call function-strings (macros, as closures...)
Metaprogramming is just string-modification and can be applied everywhere
There is just a global namespace
There are no named arguments, no function signatures
No deallocation possible
Inefficient config file
Operators are functions, but the function names are symbols
No error handling
Lack of C++ documentation (how does it work internally?)

Why this is bad:

Everything is a string means that everything is executable, which can enable an attacker to execute code in some cases. Although it is easily possible to write forkbombs and other malware for cbs, this is probably not really an horror scenario, because masterlist-providers can inject code anyway. There other scenarios, where this can be dangerous: It can lead to data(=>code,state) corruption if random data is executed (just imagine a bug where someone executes a long string of human language). Such a corrupted block might even spread…corrupted code generated more corrupted code with metaprogramming. (I call this disease cubic cancer ;) ). Also it makes debugging a lot harder and even can hide bugs, this of course goes hand in hand with error handling: No error handling => no error on code corruption.
The problems I just described could be contained a bit if there where types and function signatures anyway (line with 8 words, can not be called for function that requires 3 args; can not call a function that requires lists for a line of human language) and again…error handling.
Now there is a lot of problems with the seperation between data and code, but there is an additional problem: The usage of complicated string manipulation methods: You can use multiple Macros of different levels, macros are hard to distinguish from actual variables and their usage varys depending on the level of rect brackets. Calling a function seems to be equivalent with accessing the string one macro level deeper. Then there are two kinds of string defining symbols: [...] and "..." witch both output a string (or you can just use none of them), there are also round braces witch evaluate the code immediately. This is apperantly still a little buggy, because some strings are not recognized as entire commands and in some cases the macro modifier is not recognized (I remember, that I tried to fill a bug report about this/talk to the devs about this, I also remember that I have been somewhat unsuccessful).

The next big block of criticism is the available variable space: Ther is only the global namespace. Not a local one, no packages/namespaces, only one. In other words: You iterate over a list using I as iterator and one other function you call does the same: You are screwed. If someone else in a completely different script decides to call his var 'roster' too...you are screwed. But not only is the cbs-namespace a mess, cubescript is also one single memory leak. no RAM at all is being deallocated, try this:

testnum = 0
testmax = 2048

buildup = [ if (*smaller* $i $m) [ s = (concatword $s $s $s $s $s $s $s $s $s $s); i = (* $i 10); sleep 1 [ buildup ] ] ]
test = [
if (*smaller* $testnum $testmax) [
s = a
i = 0
m = (pow 1024 2)
buildup

sleep 1 [ test ]
]
]

The code above builds string with the size of 1M and deletes it. It does that 2048 times. The memory consumption here should not be higher than 4M (this is generous). Actually it needed about a second to fill my entire RAM.
The next thing is that the configuration is done by writing the entire state => corrupted data and functions written.
The lists are bad for the obvious reason (lets append "mapc is stupid" to the end of a list ;) ).
Operators are just badly designed, using symbols but the arithmetic of functions is inconsistent.
And at last: There are about 5 lines of comment in the entire cube engine, maybe one in the cbs code. This makes it very hard to understand the internals, even if you can read C++.

Well, I am pretty exhausted. There are some ways how you can work around this problems, I'll write them down tomorrow :) .
You said you think cbs is perfect for it's purpose, what makes you think that?
Cheers,
mapc
reply to mapc

skiingpenguins | 2011-12-15 20:38

cube script isnt supposed to be as fancy as you seem to want it. i think you should chill out, and maybe go to a different game if you want something with all these feature things. cube script serves it's purpose very well. don't mess with it.
reply to 3starskiingpenguins

mapc | 2011-12-16 01:06

@skiingpenguins

Excuse me? Do you have the faintest idea what my essay meant, have you ever spent a single thought on programming language design, have you ever thought a second about application design? Please, do not waste my time! If you did not understand what I said, that's ok. I'll even explain it to you (somewhere else), but a statement like yours is not worthy an answer and I won't feed the trolls again…
reply to mapc

Chaos | 2011-12-16 02:29

Cubescript is a low-level language; hence, it's applications don't really go beyond those for which it was designed. However, in my opinion, cubescript really servers it's purpose well, accommodating a wide range of uses that relate to the game.
reply to 1starChaos

Suicizer | 2011-12-16 08:30

? This maybe has nothing to do with your content, but CubeScript doesn't seems like a low-level programming language to me. I would rather take OpenGL as example for a low-level programming language as CubeScript.

Rating

points

(0 votes)

Comments

mapc | 2011-12-16 13:09

Again: Did you understand the subject of my critique?

Cubescript has some very bad constructs in it, the worst memory management possible (none at all) and is not very thought-out.
When you want to implement a feature in your application, you normally look first if there is a library implementing that feature (you use gzip or bzip2 for compression, you do not write your own compression method, same applies for crypto, networking…). If you do not find a lib fitting your needs, you decide whether to implement it yourself (better integration, mabe more performance, better feature, more work, more code to maintain, higher complexity) or to use a lib implementing a similar feature to what you want to do (less work, less complexity, worse feature). The question is: where is the benefit from using Cube script (instead of python or other languages)?

unixfreak | 2011-12-16 14:15

The one thing i can think of, are multiple versions of a library being used. As it can break the engine if something has drastically changed in a newer release. If you maintain your own code for a feature... it's easier to make sure it works on all systems.

As sauerbraten is pretty much a ~yearly release, during that time a library could have been updated several times, possibly breaking the engine for some users. That's usually why you would fork code or implement it yourself. I'm probably wrong in this case, but it is a valid reason for cross-platform programs.

mapc | 2011-12-16 15:20

Yes, you are right, but you still can use an old version of a library, there are even some systems, that manage different versions for you (0install for example) and Sauerbraten delivers lots of libraries with the svn, which are (I think) statically compiled into the code: Problem solved. Besides that a lot of time could be used on integrating the libs and using the libs functionality if the sb devs would not have to implement everything themselves (sauerbraten is not only the game, but also cbs, a graphics engine, a physics engine and even enet is made by eihrul. That's a lot if work.

Suicizer | 2011-12-17 11:21

Then don't ask where the benefit of CubeScript is, if that's not the point :P.

P.S.
This was actually a reply on comment #22515.

Suicizer | 2011-12-16 22:08

The benefit of CubeScript is that it's getting supported on Cube Engine 1 and 2 :). Any other programming languages than that, OpenGL/Assembly, GLSL and C++ probably aren't supported.

I'm not sure, but I always thought CubeScript already existed back in the days of Cube 1, which would mean Aardappel created CubeScript?

I think CubeScript is just like QuakeC. It's designed so you can't really modify the engine, but just change some things around it.

mapc | 2011-12-17 00:03

That is not the point... why not use zoidberg (python, pearl, lisp)

Chasester_aka_you | 2011-12-16 20:29

compared to other engine scripting languages ... it awesome but thats not saying much.

mapc | 2011-12-16 21:18

Why do you consider cubescript awesome?
Take ruby, python, pearl, groovy, js… Compared to them, cbs is pretty weak

Chasester_aka_you | 2011-12-17 01:01

-.- i said ENGINE LANGUAGES -.-

those are simi real languages

RaZgRiZ | 2011-12-17 07:52

You cannot make a game using cubescript. Can't get simpler than that.

srbs | 2011-12-17 10:11

Since this was the easiest section, I'm commenting on it and will get to the rest tomorrow.

"...cubescript is also one single memory leak. no RAM at all is being deallocated, try this:[...]
The code above builds string with the size of 1M and deletes it. It does that 2048 times. The memory consumption here should not be higher than 4M (this is generous). Actually it needed about a second to fill my entire RAM."
I would agree here, but this isn't even slightly the case. Everything when removed/lost is deallocated.

Now, look at $testnum. See how it doesn't change? Yeah, it kind of needs to, otherwise (test) will never finish.
Now, look at $i. See how it never changes? Yeah, $i should start out at 1 not 0 because 0*10=0.

Fix those two, and you'll see how wrong you are. Granted my Sauer crashed just after 1704 iterations, it still didn't run out of memory. I have no idea on the cause of the crash, but I really don't care.

srbs

Note: you can use < to get the < symbol

typhoon | 2011-12-18 00:14

It most likely crashes due to the excessive nesting. Since the "loop" here is done using sleep rather than a real loop, it calls the same function recursively over and over again. Also, I guess that contributes a good deal to the observed high memory usage.
To mapc: if you want to convince people with such a benchmark, you should at least do it properly...

mapc | 2011-12-18 19:02

Yup, your right.

I fixed and improved the script once again, and the curve of memory usage was not rising.
I suspect the vars are deallocated when changed, this however does not mean, taht there is a way of dealing with orphaned variables (like scopes).
To prevent them from taking too much space it's probably best to overwrite them with "", good to know.

// Number of tests testmax = 2048 // Test index testnum = 0


//

// Size of the string to build.

// 1byte * 1024^2 * 40 = 40MB

strsize = (* (pow 1024 2) 40)

log10strsize = (log10 $strsize)
//

// Rounds a value at the Nth decimal place (default=0 for all decimal places)

// Negative values for non-decimal places

// Args:

//    1. The value to round

//    2. The decimal places to round

//

round = [

   ml = (pow 10 $arg2) // To multiply with for decimal shift

   n = (*f $arg1 $ml)    // Shift the decimal place
   dc = (modf $n 1) // Isolate decimal places

   n = (-f $n $dc)  // Remove decimal places
   if (>=f $dc 0.5) [ // Check wether to round up or down

      n = (+f $n 1)   // If rounding up

   ]   
   n = (divf $n $ml) // Revers decimal place shift
   if (= (modf $n 1) 0)

      n = (+ $n 0)      // Remove the ".0"
   result $n

]
//

// A bit like recursion, but pushes the codeblock onto the sleep stack with a sleep time of 0.

// Elements on the sleep stack are executed during the frame whose execution point in time is &lt= the current time + the sleep time.

// For sleep times of 0 this is always the next frame (or possibly later this frame).

// This pattern can be used, to make huge computations not blaock sauerbraten, however it should be used with care, because it makes the

// control flow a lot more complicated (a bit like multithreading).

//

nextframe = [

   sleep 0 $arg1

]  
//

// Anonymous recursion utility

// It is mandatory to assign t

REC = [

   lam = $arg1

   lam [ REC [ @@lam ]]

]
//

// Build a string with the sizte of about $strsize

// Args:

//    1. The code to exec when the test has finished

//

buildup = [

   intheend = $arg1
   i = 1

   s = "a"   
   REC [

      thisblock = $arg1
      if (< $i $strsize) [

         s = (concatword $s $s $s $s $s  $s $s $s $s $s)

         i = (* $i 10)

	 log10i = (log10 $i)
         echo (concatword "Test No. " $testnum " of " $testmax ". String buildup:  1E" $log10i "/1E" $log10strsize ". Done: " (round (*f 100 (divf $testnum $testmax) ) 2) "%.")
         nextframe $thisblock

      ] [ // ELSE

         intheend

      ]

   ]

]
//

// Runs the test

//

test = [

   testnum = 0
   REC [

      thisblock = $arg1
      if (< $testnum $testmax) [

         testnum = (+ $testnum 1)

s = a i = 0 m = (pow 1024 2) buildup $thisblock ] ] ]

typhoon | 2011-12-18 00:39

I think you still didn't understand that CubeScript is NOT a programming language, and it isn't meant to be one. It's merely a language for config files, with some added scripting functionality to make your life easier.

Now let's go through some of your complaints:
"Lists are space-seperated not-escaped lists" - they aren't lists, they're strings, as you noted one line above -_-
"There is just a global namespace" - if you miss namespaces, make your own: mapc_somevarname. There's your namespace.
"There are no named arguments, no function signatures" - as I said, it's a scripting language. Did you ever use e.g. bash?
"No deallocation possible" - And none necessary. What do you need those 1MB strings for again?
"Operators are functions, but the function names are symbols" - So what? You don't have to care, as long as you remember to use prefix notation.
"No error handling" - You'll have to explain that. Do you mean Java-like exceptions or something?
"Lack of C++ documentation (how does it work internally?)" - I can't really see why you would need that, but feel free to read the code, it's quite understandable.

"Everything is a string means that everything is executable, which can enable an attacker to execute code in some cases" - well, you obviously have to be a little careful when you handle external data. But a) there's not much external data and b) it's definitely less dangerous than SQL insertion ;P

'nuff said.

The only issue I personally have with CS is that it writes even functions defined in other scripts into saved.cfg automatically. That can be annoying and/or confusing at times.

Suicizer | 2011-12-18 07:14

"It's merely a language for config files, with some added scripting functionality to make your life easier."

I hope you know that's just what I said actually on my comment of 2011-12-16 22:08?

typhoon | 2011-12-18 12:28

No, I don't.
1) I can't see any comment of yours at 2011-12-16 22:08, you'd have to tell me your time zone as well
2) I can't remember you saying something like that anyway
3) Even if you said exactly the same before, it's just a fact. My humblest apologies for repeating it...

EDIT: oh, you mean that comment where you compared CS to QuakeC? Well, for one, that's not what I said, and secondly, QuakeC is totally different, not just from the syntax and functionality, but also from how it allows you to extensively modify the game logic.

Suicizer | 2011-12-18 15:27

The timezone on quadropolis isn't my timezone, so I guess it's just the timezone where the website is getting hosted (so it should be the same for everyone).

It's very true as I haven't said it literately, but had the same meaning.

Well, QuakeC isn't that much different in purpose as CubeScript, as they both don't allow to change anything in the actual sourcecode and are both scripting languages (I guess because of that).

mapc | 2011-12-19 20:50

Cubescript is used as a programming language, therefore it is one

"they aren't lists, they're strings, as you noted one line above" - There is list support in the engine and these strings are used as such. Therefore space separated strings are lists.
"if you miss namespaces, make your own: mapc_somevarname. There's your namespace" - I do, it makes the code more complex. Besides that the namespaces should be file name dependent (uups, two "lolol" namespaces).
"as I said, it's a scripting language. Did you ever use e.g. bash?" - Write lots of scripts with it, hate it. (As soon as i need functions i use python). Most scripting languages have this feature.
"So what? You don't have to care, as long as you remember to use prefix notation." - It does not behave like polish notation, but like functions therefore it should be functions (especially for all this horrible float stuff).
"'No error handling' - You'll have to explain that. Do you mean Java-like exceptions or something?" - Yup, ore something similar.
"'Lack of C++ documentation (how does it work internally?)' - I can't really see why you would need that, but feel free to read the code, it's quite understandable." - I have a pretty hard time understanding this code and even more the params...
"you obviously have to be a little careful when you handle external data. But a) there's not much external data" - It is not only about external data, but about data anyway.
"it's definitely less dangerous than SQL insertion" - Yes, but still not good

Suicizer | 2011-12-19 20:57

"'No error handling' - You'll have to explain that. Do you mean Java-like exceptions or something?" - Yup, ore something similar.
Well, then go create something wonderful in OpenGL/Assembly, as that programming language is compatible with Cube Engine 2 and has error handling (as far as I know).

mapc | 2011-12-19 21:26

They have done it, in C/C++, it is called Python, and Ruby and JVM...

Suicizer | 2011-12-19 22:03

Yes, but unlike C++, you don't have to compile all the damn time when using OpenGL/Assembly (or even glsl), as the engine compiles it.

srbs | 2011-12-19 05:41

Cubescript is a C-style Lisp-esque hungarian notation language.
It is C in that most functions do the exact or very similar function that their C counterparts do.
It is Lisp in that it is prefix notation and the functions are editable (excluding built-in).
It is hungarian notation because, well, it's hungarian notation.

"Everything is a string (partially lack of types)"
So? In Lisp, just about everything is a list.

"Lists are space-seperated not-escaped lists"
1) You can write your own list mechanism to support escaped lists.
2) See typhoon's post.

"There are too many ways to generate and modify function-strings (e.g. macros, the three braces)"
"There are too many ways to call function-strings (macros, as closures...)"
I'm not exactly sure what you mean by these two (especially the "three braces") could you write some examples?
And yes, I did read the rest of your post that explains them, but examples are a huge help for me.

"Metaprogramming is just string-modification and can be applied everywhere"
Here I point you to metaprogramming in Perl.

"There is just a global namespace"
Yeah, It'd be nice to have local vars similar to other languages, but you can still get the same result in cubescript.

"There are no named arguments, no function signatures"
See typhoon's post.

"No deallocation possible"
See my earlier post for why this isn't true.

"Inefficient config file"
I thought we had setpersist, but it doesn't seem so. The architecture is set to make the config file more efficient, but you can always manage that yourself in cubescript.

"Operators are functions, but the function names are symbols"
Again, I will point you to Lisp.

"No error handling"
Cubescript is based on C. There is no error handling in C (at least like Java/C++/D/Python/etc).
When an error occurs, usually a -1 is returned.

"Lack of C++ documentation (how does it work internally?)"
I have had no trouble figuring out the code without any documentation.

srbs

Now don't you dare start having issues with Lisp, Perl, or Python since you used them as examples of "real" or "better" scripting languages.

mapc | 2011-12-19 21:10

"Now don't you dare start having issues with Lisp, Perl, or Python since you used them as examples of "real" or "better" scripting languages." - You really should think about your style of discussion (hint: it is not about having right)
I am going to ignore any argument that defends a feature based on it's similarity to features of other language. However, I want to note that the comment about data types in lisp is plainly wrong - (there is only one structure: list, but multiple data types) and that there is a difference between meta programming in pearl and cbs, also I do not generally disapprove of meta programming).

"It is hungarian notation because, well, it's hungarian notation." - The problem is, that it is not. fun semantic != semantic of polish notation. Btw: the notation is how the notation is, because it is the easiest thing to implement.

"I'm not exactly sure what you mean by these two (especially the "three braces") could you write some examples?
And yes, I did read the rest of your post that explains them, but examples are a huge help for me."
There are quite a few examples in the script I posted. Also you should know this stuff if you want to write cbs, but if you want, I'll be happy to show you some of the cubescript features some when on irc/mumble. Also this stuff is explaned in the cbs doc.

"Yeah, It'd be nice to have local vars similar to other languages, but you can still get the same result in cubescript." - You could also use assembler instead of cbs and you would get the things done. What matters is how you do what you do.

"You can write your own list mechanism to support escaped lists." - I could also do a remake of sauerbraten...

"but you can always manage that [config] yourself in cubescript." - Nope, I can not, that is the point (cbs does not offer I/O operations).

"No error handling" - cbs is not C++

"I have had no trouble figuring out the code without any documentation." - But I had.

Suicizer | 2011-12-19 22:07

So when you have problems with figuring out the code without any documentation, let's just blame CubeScript? That really isn't putting you in a convincing image.

mapc | 2011-12-19 22:08

@Suicizer
Do yourself a favor: Shut up before it is too late.

Chaos | 2011-12-19 23:15

Ok let's be constructive here, no need to post such a comment.

srbs | 2011-12-30 02:54

"there is only one structure [in Lisp]: list, but multiple data types"
Yeah, there is only one structure in cubescript: string, but multiple data types. It works the same way internally. Externally, (in cubescript) everything is viewed as a string. Internally, there are different types: int, float, char[].

"The problem is, that it is not. fun semantic != semantic of polish notation. Btw: the notation is how the notation is, because it is the easiest thing to implement."
I don't think you know what Hungarian notation is since it is not Polish notation and that cubescript is, in fact, Hungarian notation among others.

"There are quite a few examples in the script I posted. Also you should know this stuff if you want to write cbs, but if you want, I'll be happy to show you some of the cubescript features some when on irc/mumble. Also this stuff is explaned in the cbs doc."
I guess I'll assume what you meant.. I see no problems with multiple was of doing things since each one has a different purpose and result.

"You could also use assembler instead of cbs and you would get the things done. What matters is how you do what you do."
Ok.. you can get things done in cubescript too. I don't see your point.

"I could also do a remake of sauerbraten..."
Feel free, no one is stopping you.

"Nope, I can not, that is the point (cbs does not offer I/O operations)."
I did not say anything about I/O, especially since you cannot change the config.cfg file after written (on exit) or before read (on load). I meant that inside the scripts, functions, guis, you can manage the variables that are written to the file. I even had a almost perfect way of accomplishing this in justice & before, but in SVN that particular solution is gone.

"cbs is not C++"
No, but at it's core, it is C and error handling doesn't exist there and that extends to cubescript, unless someone feels like adding it in. But what's the point? Is there even anything that could use try/catch like handling?

"But I had."
Me too during the first two months of actually looking at the code. It's not a quick read & learn, it takes time.

srbs

RaZgRiZ | 2011-12-19 22:21

Things are getting heated and my flamewar-sense is tingling. Better keep the hose on the standby.

mapc | 2011-12-19 22:22